1. Field of the Invention
The present invention relates to a servo control system, and in particular to a servo control system that is equipped with a learning control apparatus which can optimize learning memory allocation.
2. Description of the Related Art
Learning control is known in which a servo control apparatus or a control system memorizes the relationship between a control method and its execution result and improves the control method based on the memorized result (for example, refer to Japanese Patent No. 4043996, hereinafter referred to as “patent document 1”). To implement such learning control, a memory used for learning must be secured. It is disclosed in patent document 1 that a learning control unit includes a so-called learning memory which stores correction data for one cycle of learning.
If there is no more than one axis to be controlled, one servo control unit and one learning memory will suffice for the purpose. On the other hand, if there is more than one axis to be controlled, one possible solution is to provide one learning memory for each of a plurality of servo control units. However, if, in a servo control apparatus for controlling a plurality of axes, a dedicated memory is allocated to a servo control DSP (Digital Signal Processor) for each axis, there is the problem that the cost increases correspondingly.
FIG. 1 shows a configuration example of a conventional art servo control apparatus for controlling a plurality of axes. A number, n, of servo control units (111, 112, 113, . . . ) are provided in order to control the plurality of axes, and learning control units (121, 122, 123, . . . ) and learning memories (131, 132, 133, . . . ) are provided, one each for each of the servo control units. Each learning memory is assumed to have a capacity x equivalent to maximum learning time.
There can occur a situation where some of the axes are axes (standard axes) for which learning control is not executed. In such a situation, the learning memories 132 and 133, for example, are not used for learning, and all of their memory areas remain unused; hence, there is a problem that the memories allocated to such axes are wasted.
Furthermore, the memory capacity necessary for learning varies depending on the content of the learning. However, each learning memory must be provided with a memory area sufficient to accommodate the maximum learning time. However, in this case, as in the learning memory 131, all of the memory area may be used for a learning axis A, while as in the learning memory 134, only a portion is used for a learning axis B and the remaining area remains unused. Therefore, when one learning memory is provided for each of the plurality of servo control units that control the plurality of axes, some of the memory areas remain unused, resulting in the problem that efficient use cannot be made of the learning memories. There has also been the problem that it is not possible to execute learning that takes longer than when the memory area of each of the learning memories provided in the respective servo control units can be accommodated.